Google Content Experiments is a tool that allows you to try varieties of content to find the solution that converts the best.
There are a variety of experiments to try. For example, If you are testing an eCommerce site, you could test what variation helps with the conversion of a user purchase. Some of the most common goals to test are:
- User purchases and overall conversion value
- Subscription list sign ups
- Overall revenue
- Session duration
- Bounce rate
- Landing page design
Benefits of these experiments
Google Content Experiments will give you valuable insight into how to improve your site. Sometimes you have to find what isn’t working to figure out what IS working. These experiments will help you with:
- Traffic assessment.
- Comparing how different web pages perform by using a random sample of your users.
- Choosing which objective you’d like to focus on, i.e. more conversions, more time spent on site, more inquiries.
Let’s pretend that you’re hired to improve a website for a landscape company. You’re in charge of the content for the various business offerings: landscape design, retaining walls, outdoor kitchens, and irrigation systems. The website is the primary sales tool for potential customers to get in touch with the business. Irrigation systems are what the business would like to become better known for and start gaining market share. The goal is to get users to convert so the landscape company can provide this service to their clients while getting more prospects interested.
It’s a relatively small business and according to your analytics, most users land on your homepage, so this is a good thing to test first. When testing, several new versions of this homepage should be created. For example, one could be more imagery heavy, one with a large headline, and one could be shown with a special offer as the main focus.
Once the experiment is up and running, a random sample of your website visitors will see each version. Whatever variation provides the results you like best, in this case inquiring about irrigation systems, is the one that should be the chosen variation going forward.
What you need in order to create an experiment
There are plenty of A/B testing tools out there, which are often subscription based. Although they offer some great features, you may not necessarily need a separate tool for your testing, especially if you’re just getting started. The good news is, there is no cost to use Google Content Experiments. With this free solution, all you need is a Google Analytics account. The experiments are set up through the Google Analytics dashboard, which is great because then all of your data is in one place.
If you do not have Google Analytics installed, it’s just a matter of adding a script to your website. This unique code is tied to your website and you can gather all sorts of valuable data from it. For the experiment, there is some additional code that you will need, but is easily generated and implemented. With WordPress, there are a few different options for how to add Google Analytics to your WordPress site.
How to set up an experiment
We will be testing landing pages and using a few different variations. The idea here is to test which version of a landing page results in the greatest improvement in conversions. If you’re really serious about testing, you can test up to ten variations of a landing page.
If you go to the left-hand side of the Google Analytics dashboard, you will see a list of options. Go to Behavior > Experiments. Click on the “create experiment” button to get started. Here you will notice a spot for your experiment list. You will see other experiments if any have been previously set up.
The experiment requires the same domain for the different variations. For example mywebsite.com and thisismywebsite.com cannot be part of the same experiment. However, www.mywebsite.com/variation-1 and www.mywebsite.com/variation-2 share the same domain, so this will work for the experiment.
Step one: Choose an experiment objective
This is where you will name your experiment. If you are running multiple, giving descriptive names will be helpful. For the objective of this experiment, you can choose bounces, page views, session duration, or create a new objective. This is the metric that the experiment is evaluated with and how the winner will be chosen.
Under “Percentage of traffic to experiment,” choose the percentage of users that you’d like to include in the experiment. It’s important to consider that the original page is part of the experiment. If you only choose 50% of your traffic with two variations, only 25% of the traffic has the chance of seeing the new variation; 75% will see the original. This is OK, but it will take longer for the experiment to run.
It’s helpful to receive email notifications on the status of the experiment, so be sure to turn that feature on.
There are advanced options that aren’t absolutely necessary but may provide more insight to your experiment. If you choose to distribute traffic evenly across all variations, there will be an equal amount of traffic to each variation for the life of the experiment. If this is disabled, the default behavior is to dynamically adjust traffic based on variation performance.
Setting a minimum time that the experiment will run is the option to provide the minimum period where Google Analytics will not declare a winner. Let’s say your content is seasonal, or has behavior patterns depending on the time of day, this might be a good option to make sure that the winner isn’t determined too quickly.
When you set the confidence threshold, keep in mind that the higher the threshold, the more confident you can be in the result. Higher thresholds also mean that the experiments will take longer to have a determined winner.
Step two: Configure your experiment
This is where things start to get real. From here, we will be able to set and preview the different variations. First, you will enter the URL for the original page. A preview should show up right next to it. Next, you will enter the URLs for the variations. If you have more than a few, having a detailed name for them is beneficial.
Step three: Setting up your experiment code
We have our Google Analytics code, but we need a code snippet that is specific to this experiment. This is generated for you and will need to be added to your site. One thing to note, it has to go before the code for Google Analytics. When the snippet is placed on your site, you will receive a message that the experiment can be tracked.
Step four: Review and start
Do a quick double check to make sure that the code is working and there is a check mark next to each experiment. When all looks good, all you have to do is add some optional notes and click the button to start the experiment. Depending on your traffic levels, it may take some time to see any results.
How to interpret experimental results
The results are pretty self-explanatory. Be sure to check out the experiment list which contains general information about all your experiments. It’s a great reference for tracking the status and associated traffic. You’ll see the status of each experiment: Setup, Running, Stopped, or Ended and the total number of visits to pages in each experiment.
If you’re familiar with Google Analytics, the charting styles on the report will be very familiar. Colored lines in the chart indicate variation and you can compare that with the original. You can choose the timeframe and get detailed information about how the variations performed.
As Google Analytics continues to be one of the most popular analytics tools, using a simple setup will allow you to be up and running website experiments in no time. It doesn’t have to be expensive to set up experiments like this, and the efforts can have a huge impact on your business.